STE20-type kinase TAOK3 regulates hepatic lipid partitioning

Objective Nonalcoholic fatty liver disease (NAFLD), defined by excessive lipid storage in hepatocytes, has recently emerged as a leading global cause of chronic liver disease. The aim of this study was to examine the role of STE20-type protein kinase TAOK3, which has previously been shown to associate with hepatic lipid droplets, in the initiation and aggravation of human NAFLD. Methods The correlation between TAOK3 mRNA expression and the severity of NAFLD was investigated in liver biopsies from 62 individuals. In immortalized human hepatocytes, intracellular fat deposition, lipid metabolism, and oxidative and endoplasmic reticulum stress were analyzed when TAOK3 was overexpressed or knocked down by small interfering RNA. Subcellular localization of TAOK3 was characterized in human and mouse hepatocytes by immunofluorescence microscopy. Results We found that the TAOK3 transcript levels in human liver biopsies were positively correlated with the key lesions of NAFLD (i.e., hepatic steatosis, inflammation, and ballooning). Overexpression of TAOK3 in cultured human hepatocytes exacerbated lipid storage by inhibiting β-oxidation and triacylglycerol secretion while enhancing lipid synthesis. Conversely, silencing of TAOK3 attenuated lipid deposition in human hepatocytes by stimulating mitochondrial fatty acid oxidation and triacylglycerol efflux while suppressing lipogenesis. We also found aggravated or decreased oxidative/endoplasmic reticulum stress in human hepatocytes with increased or reduced TAOK3 levels, respectively. The subcellular localization of TAOK3 in human and mouse hepatocytes was confined to intracellular lipid droplets. Conclusions This study provides the first evidence that hepatic lipid droplet-coating kinase TAOK3 is a critical regulatory node controlling liver lipotoxicity and susceptibility to NAFLD.


INTRODUCTION
Nonalcoholic fatty liver disease (NAFLD) is defined as lipid accumulation in >5% of hepatocytes (steatosis) in the absence of excessive alcohol use (!30 g per day for men and !20 g per day for women) [1].
Recently, NAFLD has emerged as a leading global cause of chronic liver disease due to the exponential rise in obesity, which is the main risk factor for the development and aggravation of NAFLD [2]. Importantly, about 10e20% of patients with NAFLD progress to nonalcoholic steatohepatitis (NASH), which is characterized by local inflammation and cellular injury (ballooning) in the liver, in addition to fat infiltration [3]. Patients with NASH are at high risk of developing cirrhosis, liver failure, and hepatocellular carcinoma (HCC), which is one of the most fatal and fastest-growing cancers [4]. Thus, understanding the molecular pathogenesis of NAFLD is of high clinical importance for the efficient prevention and management of a range of complex liver diseases.
In NAFLD, lipids accumulate within intrahepatocellular lipid droplets composed of a neutral core of triacylglycerol (TAG) and cholesterol esters, surrounded by a phospholipid monolayer that harbors a unique set of proteins [5]. Liver lipid droplet-associated proteins are increasingly recognized as critical regulators of not only hepatic lipid metabolism but also protein quality control and storage, cell signaling, viral replication, and interactions with other organelles [6]. Notably, among genetic variants that confer susceptibility to NAFLD, the best characterized are a single-nucleotide polymorphism in the PNPLA3 gene and a splice variant (rs72613567:TA) in the HSD17B13 gene, both of which encode proteins anchored to the liver lipid droplets [7,8]. Furthermore, we have recently demonstrated that several STE20-type kinases e STK25, MST3, and MST4 e associate with intrahepatocellular lipid droplets and critically orchestrate liver lipid partitioning and NAFLD development [9e17]. Consequently, mapping the composition of the lipid droplet proteome in hepatocytes and exploring its mode of action in the control of liver lipid homeostasis are essential for deciphering the molecular pathophysiology of NAFLD. Our recent studies using global proteomic analysis have identified thousand and one kinase 3 (TAOK3; also known as MAP3K18, JIK, or DPK) as a lipid droplet-associated protein in mouse liver [14,15]. TAOK3 is a STE20-type kinase, widely expressed in different cell types and previously linked to the regulation of MAPK signaling [18,19]. The role of TAOK3 in the immune system is emerging as it has recently been reported to regulate the commitment to the marginal zone B cell fate [20], terminal differentiation of conventional dendritic cells [21], and canonical TCR signaling [22]. The role of TAOK3 has also been implicated in tumor initiation and metastasis formation in pancreatic cancer and reduced cell death in breast cancer [18,23]. However, the function or mode of action of this kinase in hepatocytes has not been investigated yet. In this study, we provide several lines of evidence suggesting a possible role of TAOK3 in the initiation and aggravation of human NAFLD. We found that TAOK3 mRNA levels in human liver biopsies correlate positively with the severity of NAFLD. Furthermore, we show that the overexpression or silencing of TAOK3 in human hepatocytes results in exacerbated or reduced, respectively, lipid accumulation as well as oxidative and endoplasmic reticulum (ER) stress. In line with our previous observations in mouse liver [14,15], we found that the subcellular localization of TAOK3 is confined to intracellular lipid droplets in human hepatocytes.

Analysis of the TAOK3 mRNA expression in human liver biopsies
The TAOK3 mRNA expression was analyzed in interoperative liver biopsies collected from Caucasian subjects (men, n ¼ 35; women, n ¼ 27) who underwent laparoscopic abdominal surgery for Roux-en-Y bypass (n ¼ 12), sleeve gastrectomy (n ¼ 9), or elective cholecystectomy (n ¼ 41). The participants fulfilled the following inclusion criteria: (1) men and women aged >18 years; (2) indication for elective laparoscopic or open abdominal surgery; (3) BMI between 18 and 50 kg/m 2 ; (4) abdominal MRI feasible; and (5) signed written informed consent. The exclusion criteria for liver biopsy donors were as follows: (1) significant acute or chronic inflammatory disease or clinical signs of infection; (2) CrP >10 mg/dl; (3) type 1 diabetes and/or antibodies against glutamic acid decarboxylase (GAD) and islet cell antibodies (ICA); (4) systolic blood pressure >140 mmHg and diastolic blood pressure >95 mmHg; (5) clinical evidence of cardiovascular or peripheral artery disease; (6) thyroid dysfunction; (7) alcohol or drug abuse; and (8) pregnancy. Type 2 diabetes (T2D) was diagnosed by a fasting plasma glucose value >7.0 mmol/l and/or a 120-min oral glucose tolerance test (OGTT) glucose value >11.1 mmol/l. For participant characteristics, refer to Cansby et al. [15]. Body fat was analyzed by dual X-ray absorptiometry (DEXA) and liver fat was measured by single-proton magnetic resonance spectroscopy ( 1 H-MRS) as described previously [24]. A small liver biopsy was obtained during the surgery (between 08:00 and 10:00 h after overnight fasting), immediately snap frozen in liquid nitrogen, and stored at À80 C. The NAFLD activity score (NAS) and fibrosis score were assessed on liver sections by a certified pathologist [25]. qRT-PCR was performed in liver biopsies as described below using the probes for TAOK3 (Hs00937694_m1) and 18S rRNA (Hs99999901_s1; Thermo Fisher Scientific, Waltham, MA), which span exoneexon boundaries to improve the specificity. All subjects gave written informed consent to use their data in anonymized form for research purposes before taking part in this study. All investigations were approved by the Ethics Committee of the University of Leipzig, Germany (approval numbers 363-10-13,122,010 and 159-12-21,052,012), and were performed in accordance with the Declaration of Helsinki.  [14]. In parallel, IHHs were processed for immunofluorescence with anti-TAOK3, anti-adipose differentiation-related protein (ADRP), anti-8-oxoguanine (8-oxoG), anti-4-hydroxynonenal (4-HNE), anti-E06, anti-KDEL, anti-C/EBP-homologous protein (CHOP), antiperoxisomal biogenesis factor 5 (PEX5), or anti-peroxisomal membrane protein 70 kDa (PMP70) antibodies (see Supplementary Table S1 for antibody information). Immunofluorescence images were acquired using a Zeiss Axio Observer microscope with the ZEN Blue software (Zeiss, Oberkochen, Germany). The labeled area was quantified in 6 randomly selected microscopic fields (Â20) per well using the ImageJ software (1.47v; National Institutes of Health, Bethesda, MD). Membrane lipids were extracted using the BUME method [27]; free cholesterol was quantified using straight-phase HPLC with an evaporative light scattering detector (ELD) as previously described [28], while sphingomyelin, phosphatidylcholine, lysophosphatidylcholine, and phosphatidylethanolamine were measured using direct infusion on a QTRAP 5500 mass spectrometer (Sciex, Concord, Canada) equipped with a robotic nanoflow ion source (TriVersa NanoMate; Advion Bio-Sciences, Ithaca, NJ) [29].
To measure b-oxidation, IHHs were incubated in the presence of (9,10-3 H[N])palmitic acid (PerkinElmer, Waltham, MA), and [ 3 H]labeled water was quantified as the product of free fatty acid oxidation [11]. TAG secretion and incorporation of [ 14 C]glucose (PerkinElmer) into TAGs were assessed as described previously [9]. Fatty acid uptake was measured using the Quencher-Based Technology (QBT) Fatty Acid Uptake Assay Kit (Molecular Devices, San Jose, CA). Cell viability was detected using the CellTiter-Blue Cell Viability Assay (Promega, Stockholm, Sweden) according to the manufacturer's recommendations. The TAG hydrolase activity was determined in total cell lysates using [ 3 H]triolein (PerkinElmer) as the substrate [17].
Targeted metabolomics was carried out to analyze phosphohexoses and amino acids by multiple reaction monitoring scan on a QTRAP 4500 mass spectrometer (Sciex, Concord, Canada) as described previously [30]. . The relative quantities of the target transcripts were calculated after normalization of the data to the endogenous control, 18S rRNA (Thermo Fisher Scientific). Coimmunoprecipitation was carried out using anti-MYC antibodies according to the manufacturer's instructions (Anti-c-MYC Magnetic Beads; Thermo Fisher Scientific). Western blot analysis was performed as described previously [31] (see Supplementary Table S1 for antibody information).

Yeast two-hybrid analysis
Yeast two-hybrid (Y2H) screening was performed by Hybrigenics Services, S.A.S., Evry, France (http://www.hybrigenics-services. com) as described below. The coding sequence for Homo sapiens STK25 (NM_006374.5; positions 181 to 1461) was PCR-amplified and cloned into pB27 as a C-terminal fusion to the LexA DNAbinding domain (LexA-STK25) and into pB35 as a C-terminal fusion to the Gal4 DNA-binding domain (Gal4-STK25). The constructs were checked by sequencing and used as a bait to screen a random-primed primary human hepatocyte cDNA library constructed into pP6. pB27 and pP6 were derived from the original pBTM116 [32] and Pgadgh [33] plasmids, respectively. pB35 was constructed by inserting the Gal4 DNA-binding domain from pAS2DD [34] into the pFL39 backbone [35] under the control of the MET25 promoter [36].
For the LexA-STK25 bait construct, 66 million clones (5.5-fold the complexity of the library) were screened using a mating approach with YHGX13 (Y187 ade2-101::loxP-kanMX-loxP, mata) and L40DGal4 (mata) yeast strains as described previously [34]. 35 Hisþ colonies were selected on a medium lacking tryptophan, leucine, and histidine. For the Gal4-STK25 bait construct, a total of 74 million clones (6.1-fold the complexity of the library) were screened using the same mating approach with YHGX13 (mata) and CG1945 (mata) yeast strains. 81 Hisþ colonies were selected on a medium lacking tryptophan, leucine, methionine, and histidine. The prey fragments of the positive clones were amplified by PCR and sequenced at their 5 0 and 3' junctions. The resulting sequences were used to identify the corresponding interacting proteins in the GenBank database (NCBI) using a fully automated procedure.
A confidence score (PBS, for predicted biological score) was attributed to each interaction as described previously [37]. Briefly, the PBS relies on two different levels of analysis. First, a local score takes into account the redundancy and independency of the prey fragments, as well as the distribution of reading frames and stop codons in the overlapping fragments. Second, a global score takes into account the interactions found in all the screens performed at Hybrigenics using the same library. This global score represents the probability of an interaction being nonspecific. For practical use, the PBS scores were divided into four categories, from A (highest confidence) to D (lowest confidence). A fifth category (E) specifically flags interactions involving highly connected prey domains previously found several times in screens performed on libraries derived from the same organism. Finally, several of these highly connected domains have been confirmed as false-positives of the technique and are tagged as F. The PBS scores have been shown to positively correlate with the biological significance of the interactions [38,39].

Statistical analysis
The statistical significance between the groups was evaluated using the two-sample Student's t-test with a value of P < 0.05 considered statistically significant. The correlation between the TAOK3 expression in human liver biopsies and the hepatic lipid content, NAS, and fibrosis score was investigated by Spearman's rank correlation analysis after the assessment of normality of data using the KolmogoroveSmirnov test. All statistical analyses were conducted using SPSS statistics (v27; IBM Corporation, Armonk, NY).

RESULTS
3.1. TAOK3 expression in the human liver is positively correlated with the severity of NAFLD NAFLD is defined by the excessive accumulation of fat in the liver [1]. Thus, we first analyzed the correlation between the expression of TAOK3 mRNA in liver biopsies and the hepatic fat content measured by magnetic resonance spectroscopy ( 1 H-MRS) in a cohort of 62 participants representing a wide range of BMI (22.7e45.6 kg/m 2 ) and body and liver fat content (19.5e57.9% and 1.1e50.0%, respectively). We found that the hepatic TAOK3 mRNA abundance was positively correlated with the liver fat levels ( Figure 1A). Notably, there was no correlation between the TAOK3 transcript and the gender, BMI, body fat, waist-to-hip ratio, or HbA1c values of the subjects (Supplementary Figure S1). Next, we evaluated the association between hepatic TAOK3 mRNA and the widely used histological score of NAFLD severity e NAS. We found a positive correlation between the TAOK3 levels and all three individual components of the NAS (i.e., liver steatosis, lobular inflammation, and hepatocellular ballooning) as well as total NAS ( Figure 1BeE). Notably, subjects with NAS !5, which defines definite NASH (n ¼ 24), had a 1.8 AE 0.2-fold increase in the TAOK3 expression compared with subjects with NAS 4, which indicates simple steatosis or borderline NASH (n ¼ 38; P ¼ 0.008). Importantly, a significant positive correlation was also detected between TAOK3 abundance and the histological fibrosis score ( Figure 1F). It has been well established that NAFLD and T2D coexist and act synergistically to cause negative outcomes in clinical practice [1].
Based on this evidence, an additional correlation analysis was performed in a subset of 24 participants who had been diagnosed with T2D. Significant positive correlations were found between the hepatic TAOK3 mRNA levels and the liver fat content, as well as NAS and the fibrosis score, even in this cohort (Supplementary Figure S2).
Recently, NASH has been recognized as a major catalyst for HCC, which is one of the most fatal and fastest-growing cancers [40e42]. Interestingly, the microarray data analysis of the large cohort of HCC patients available at the GEO database (n ¼ 214) demonstrated that the TAOK3 gene expression was significantly higher in HCC tumors than in the adjacent non-tumor liver tissue (P < 0.0001; GSE14520). TAOK3 protein abundance was also enhanced in frequently used human HCC cell lines (poorly differentiated SNU-475 and welldifferentiated Hep3B) versus non-tumor IHHs (Supplementary Figure S3). Furthermore, TAOK3 protein levels were increased in the livers of mice with HCC induced by the administration of diethylnitrosamine (DEN) combined with a high-fat diet or treatment with carbon tetrachloride (CCl 4 ) combined with a choline-deficient L-aminoacid-defined (CDAA) diet, when compared with those observed in the samples collected from healthy chow-fed control mice (Supplementary Figure S4).

TAOK3 decorates intrahepatocellular lipid droplets
Our previous studies have identified TAOK3 by proteomic analysis performed on the lipid droplet fraction from steatotic livers of high-fatdiet-fed mice [14,15]. Importantly, the proteins isolated by this approach do not necessarily represent bona fide lipid droplet proteins residing primarily or exclusively on the droplets, since the method fails to effectively deplete the membrane-bound cellular organelles that closely associate with lipid droplets, including the ER, mitochondria, peroxisomes, and endosomes [43,44]. Here, we further investigated Original Article the subcellular localization of TAOK3 in human hepatocytes and mouse liver using immunofluorescence microscopy. We found that TAOK3 protein was closely associated with lipid droplets, visualized by ADRP (also known as adipophilin or perilipin-2) staining, both in cultured human hepatocytes and in mouse liver sections (Figure 2A). Interestingly, we also found that hepatic TAOK3 protein abundance was increased 3.8 AE 0.6-fold in mice in response to the challenge with a high-fat diet ( Figure 2B). Notably, we detected the TAOK3 transcript in all the human, mouse, and rat tissues analyzed ( Figure 2C), which is consistent with the findings of previous studies describing the ubiquitous expression pattern of human TAOK3 [20,45]. The high TAOK3 levels observed in organs that do not store a significant amount of lipids, particularly in the human lungs, suggest that TAOK3 may display a different subcellular localization pattern in extrahepatic tissues.

TAOK3 controls intrahepatocellular lipid partitioning
To further examine the causative relationship between hepatic TAOK3 levels and liver steatosis, we characterized the effect of modifying the TAOK3 abundance on lipid metabolism in human hepatocytes. For TAOK3 overexpression, we transfected IHHs with human MYC-tagged TAOK3 expression plasmid or an empty control plasmid (Supplementary Figure S5). In parallel to the assays performed under basal culture conditions, we also treated the cells with oleic acid to replicate the environment in high-risk subjects (Supplementary Figure S5). IHHs transfected with MYC-tagged TAOK3 displayed a robust increase in TAOK3 mRNA and protein abundance compared with cells transfected with vector control (Figure 3A and B). Immunostainings with anti-MYC antibody also demonstrated a high transfection efficacy of >90% in IHHs transfected with MYC-TAOK3 expression plasmid (Supplementary Figure S6). First, we stained the transfected cells with Bodipy 493/503 for visualizing neutral lipids within the lipid droplets. We found a significant increase in the Bodipystained area in TAOK3-overexpressing IHHs ( Figure 3C  The reason for this discrepancy remains unknown. TAOK3 overexpression also significantly decreased the secretion of de novosynthesized TAG into the media ( Figure 3F). Reciprocally, there was a 2-fold increase in the incorporation of medium-derived glucose into intracellular TAG in IHHs transfected with the TAOK3 expression plasmid versus vector control, with no alteration in the fatty acid uptake ( Figure 3G and H). The concentrations of phosphohexoses and several amino acids were significantly lower in TAOK3-overexpressing hepatocytes, which likely reflects a shift in the energy metabolism from the utilization of lipids toward carbohydrates and amino acids (Supplementary Figure S9A). Importantly, increased TAOK3 abundance had no effect on the cell viability ( Figure 3I).
To study the metabolic effects of TAOK3 silencing in human hepatocytes, we transfected IHHs with TAOK3-specific siRNA or with a nontargeting control (NTC) siRNA (Supplementary Figure S10). As expected, TAOK3 mRNA and protein expression was efficiently silenced in cells transfected with TAOK3 siRNA (Figure 4A and B). In contrast to our observations in TAOK3-overexpressing cells, the Bodipy-positive area was significantly lower in TAOK3-deficient IHHs ( Figure 4C and D). Noteworthily, this difference in the cellular lipid deposition did not affect the membrane lipid composition (Supplementary Figure S7B). We also found that TAOK3 knockdown in IHHs enhanced mitochondrial biogenesis, fatty acid oxidation, and the rate of TAG secretion (Figure 4CeF; Supplementary Figure S8B). In contrast, both de novo lipogenesis and fatty acid influx were suppressed in IHHs where TAOK3 was silenced ( Figure 4G and H). Notably, the significant increase in boxidation and the reduction in free fatty acid uptake were detected only in TAOK3-deficient cells incubated with oleic acid (Figure 4E and H). The concentration of phosphohexoses and the levels of most of the amino acids were unaffected by TAOK3 knockdown (Supplementary Figure S9B). We observed no difference in cell viability in IHHs transfected with TAOK3 siRNA versus NTC siRNA ( Figure 4I). Based on the close association of TAOK3 with intrahepatocellular lipid droplets, we next tested the hypothesis that TAOK3 controls lipid mobilization from the droplets by regulating canonical lipolysis. In this process, a series of lipases binding to the lipid droplet surface sequentially catabolize TAG into free fatty acids, which are subsequently targeted to mitochondria for b-oxidation or directed to the ER/ Golgi for the synthesis and secretion of very low-density lipoprotein (VLDL)-TAG [46]. Indeed, we detected significantly reduced or enhanced lipolysis rate (assay measuring TAG hydrolase activity using triolein as the substrate [47]) in IHHs treated with oleic acid, when TAOK3 was overexpressed or knocked down, respectively ( Figure 5A and B).
As an alternative to lipolysis, lipids can be mobilized from the droplets by selective autophagy (also called lipophagy) [48]. To test the hypothesis that TAOK3 regulates lipid catabolism via autophagy, we measured the conversion of LC3-I to LC3-II, which is considered to be a key marker of autophagic flux. However, we did not detect any alterations in the LC3-II to LC3-I ratio in IHHs where TAOK3 was overexpressed or silenced ( Figure 5C and D).

TAOK3 regulates oxidative and endoplasmic reticulum stress in human hepatocytes
Accumulation of excessive lipids in hepatocytes is known to aggravate oxidative and ER stress, which in turn fuels liver inflammation, fibrosis, and apoptosis, thus triggering disease progression from simple steatosis toward NASH [49]. Consistent with this evidence, we observed that exacerbated lipid storage in TAOK3-overexpressing IHHs was accompanied by significantly increased oxidative stress as evidenced by the higher abundance of superoxide radicals (O À ) measured by DHE staining, elevated oxidative DNA damage quantified by immunostaining for 8-oxoG, and increased deposition of lipid peroxidation products and oxidized phospholipids detected by immunostaining for 4-HNE and E06, respectively ( Figure 6A; Supplementary Figure S11A). Furthermore, we found enhanced immunostaining for KDEL (a signal motif for ER retrieval) and CHOP (an indicator of ER stress-induced cell death) in IHHs transfected with the MYC-tagged TAOK3 expression plasmid versus vector control ( Figure 6A; Supplementary Figure S11A Figure S11B). At present, we cannot explain the mechanisms by which TAOK3 regulates transcription. However, a lower transcriptional response in TOAK3-overexpressing IHHs suggests a possibility that different critical thresholds for TAOK3 may exist in hepatocytes. Interestingly, we found increased or diminished peroxisomal activity in IHHs in which TAOK3 was overexpressed or silenced, respectively, as shown by alternations in immunostaining for the peroxisome biogenesis marker PEX5 and the peroxisomal membrane protein PMP70 ( Figures 6A and 7A; Supplementary Figures S11A and S12A).
3.5. TAOK3 deficiency inhibits JNK signaling and suppresses the STE20-type kinase STK25 abundance Silencing of TAOK3 has previously been implicated both in the inhibition (human mesenchymal stromal cells) and activation (HeLa cervical carcinoma cells) of c-Jun-N terminal kinase (JNK) signaling [50,51]; however, the possible role of TAOK3 in the regulation of the JNK pathway has not been studied in hepatocytes. Here we found that the phosphorylation of JNK was significantly lower in IHHs transfected with TAOK3 siRNA compared with NTC siRNA, both under basal culture conditions and after exposing the cells to oleic acid ( Figure 8A). TAO kinases have also been implicated in the regulation of p38 MAPK and Hippo pathways [45]. We did not detect any significant alterations in the phosphorylation of Yesassociated protein (YAP), the major effector of Hippo signaling, in TAOK3-deficient hepatocytes ( Figure 8A) while phosphorylation of p38 remained below the level of quantification in IHHs transfected with both TAOK3 siRNA and NTC siRNA (data not shown).
Our recent studies have revealed that depletion of STE20 kinases STK25, MST3, or MST4 protects against intrahepatocellular lipid accumulation [9e17], similarly to the silencing of TAOK3. In the light of this evidence, we next investigated whether the abundance of STK25, MST3, or MST4 protein was altered in TAOK3-deficient hepatocytes. We found that STK25 levels were significantly lower in IHHs transfected with TAOK3 siRNA versus NTC siRNA, while no difference was detected in MST3 or MST4 ( Figure 8A). TAOK3 appears to regulate STK25 abundance at the level of protein expression and/or stability, since the STK25 mRNA content was not affected (Supplementary Figure S13). To identify the novel interacting proteins of STK25, we performed a genome-wide Y2H screening of a primary human hepatocyte cDNA library using full-length human STK25 fused to the LexA or Gal4 DNAbinding domain as bait. The screening revealed a total of 11 potential interaction partners of STK25 ( Figure 8B). Interestingly, TAOK3 was identified as the binding partner for STK25 when using the Gal4-STK25 as bait, which is considered to have higher sensitivity for detecting weak interactions compared with LexA fusions. The prey fragment of TAOK3 that interacted with STK25 (amino acids 1 to 447) harbors the N-terminal kinase domain and the serine-rich domain of unknown function but excludes the C-terminal coiled-coil regions implicated in the formation of TAOK3 oligomers [45,52]. Notably, we recovered clones for the previously identified STK25-interactors GOLGA2 (also known as GM130 or GOLGIN A2) [53,54], PDCD10 (also known as CCM3 or TFAR15) [55,56], and CAB39 (also known as MO25-a) [57], verifying the high quality of the screen.
To confirm the direct interaction between TAOK3 and STK25, we next transfected IHHs with plasmids encoding MYC-TAOK3 and FLAG-STK25. Using anti-MYC immunoprecipitation, we were able to verify the direct binding between full-length TAOK3 and STK25 ( Figure 8C). Furthermore, by immunoprecipitation of the protein extracts from IHHs transfected with the truncated versions of the MYC-tagged TAOK3 expression plasmid, we confirmed that the N-terminal fragment of TAOK3-containing kinase and serine-rich domains, but not the Cterminal fragment of the TAOK3-containing coiled-coil regions, interacted with STK25 (Supplementary Figure S14).

DISCUSSION
Deciphering the molecular mechanisms underlying the onset and progression of NAFLD is essential to discover effective therapeutic strategies for its prevention and treatment. This study provides the first evidence for a possible role of the liver lipid droplet-binding STE20type kinase TAOK3 in NAFLD pathogenesis. The primary driver in NAFLD is an imbalance in hepatic lipid metabolism, which leads to the accumulation of intrahepatocellular fat, which then fuels oxidative and ER stress, local inflammation, cell damage, and subsequent fibrogenesis in the liver [49,58]. Importantly, we found that the overexpression of TAOK3 in human hepatocytes enhanced de novo lipogenesis (i.e., input) and reduced b-oxidation and TAG secretion (i.e., output) resulting in aggravated fat storage within lipid droplets, and the opposite effect was observed in TAOK3-deficient hepatocytes ( Figure 8D). Consistently, oxidative and ER stress was substantially exacerbated or suppressed in human hepatocytes where TAOK3 was overexpressed or silenced, respectively ( Figure 8D). In line with these results, we found that TAOK3 expression in human liver biopsies was positively correlated with hepatic steatosis measured by magnetic resonance spectroscopy as well as histological scoring. Furthermore, we detected a significant positive correlation between TAOK3 levels in human liver biopsies and hepatic inflammation, cell damage, and fibrosis scores. Together, these results suggest that TAOK3 regulates a Original Article 10 critical node governing hepatocellular lipid homeostasis and that TAOK3 antagonism could mitigate NAFLD initiation as well as disease progression toward NASH.
The main technical limitation of this study is our inability to delineate the primary versus the secondary changes in response to the modifications in the abundance of TAOK3 in hepatocytes. For example, alterations in oxidative and ER stress can be caused by increased or decreased lipid storage in cells where TAOK3 is overexpressed or silenced, respectively; however, TAOK3 may also have a direct impact on the metabolic stress response.
Notably, we found that TAOK3 displays a very distinct subcellular localization pattern in both human and rodent hepatocytes and is associated with intracellular lipid droplets. We also demonstrate that TAOK3 protein abundance was increased severalfold in mouse liver in response to a challenge with a high-fat diet. This observation is in line with the recent evidence showing that the hepatic lipid droplet  proteome is highly dynamic, being influenced by dietary composition as well as the liver metabolic status [46]. Interestingly, we observed that the silencing of TAOK3 suppressed JNK signaling in hepatocytes. Recent reports have revealed that decreased mitochondrial fat oxidation in liver steatosis and NASH is initiated by the activation of JNK, which can phosphorylate mitochondrial proteins [59]. Reciprocally, hepatocyte-specific inhibition of the JNK pathway has been demonstrated to enlarge the mitochondria and increase mitochondrial b-oxidation, protecting the mice from liver steatosis [60]. Thus, decreased JNK phosphorylation may have contributed to a significant increase in the mitochondrial activity observed in TAOK3deficient hepatocytes. We found that TAOK3 forms a complex in human hepatocytes with another kinase of the STE20 family e STK25. Furthermore, the silencing of TAOK3 in hepatocytes significantly reduced the protein abundance of STK25. This finding is relevant in light of our recent studies demonstrating that depletion of STK25, similarly to TAOK3 deficiency, suppresses intracellular fat accumulation in hepatocytes by shifting the metabolic balance from lipid synthesis toward lipid utilization [9e14]. Thus, the regulation of STK25 levels may constitute a part of the mechanism of action of TAOK3 in the liver. In addition to TAOK3 and STK25, the STE20-type kinases MST3 and MST4 have been associated with molecular pathogenesis of NAFLD, as evidenced by (i) protein localization to the surface of intrahepatocellular lipid droplets, (ii) inhibition of the target suppressing liver lipid accumulation in mouse models and/or cultured human hepatocytes, (iii) a significant correlation between the hepatic expression levels of the kinase and the severity of NAFLD in humans [15e17]. At this juncture, we are only in the infancy of mechanistic understanding of the role of STE20 kinases in liver lipid droplet dynamics and NAFLD; however, this family emerges as an important node regulating hepatic lipotoxicity, warranting further investigations in basic biology as well as clinical implications. NAFLD is currently the most rapidly increasing cause of liver-related morbidity and mortality, with a substantial health economic burden and no approved therapy [1]. The primary insult in NAFLD is excessive fat accumulation within hepatic lipid droplets and, therefore, the therapeutic strategies against NAFLD may include protection against and/or resolution of liver steatosis, analogous to the concept of lowering lipids to prevent atherosclerosis. Our study emphasizes the importance of lipid droplet-binding proteins in the control of intrahepatocellular fat storage and highlights the inhibition of lipid droplet decorating STE20-type kinase TAOK3 as a potential NAFLD therapy via antagonizing hepatic steatosis.